 Ok, so starting the presentation about Hermes, that's our project related to using HF as backhaul for communities in very isolated places like Amazon Forest or Central Africa, things like this. So first, why the HF band? Because the ionosphere that ranges more or less between 50 to 250 kilometers of height, it can bounce the signals that you transmit in this band during the time you want. There are the more important layers that reflect the signal is this F1 and F2. During the day there exist two layers, then they merge together at night. In the Ecuador, close to the Ecuador line, there is another stratification that the F2 stratifies to F2 and F3. Depending on the literature, people say 2.1 or F3. What determines how good your signal will bounce in the ionosphere is the density of the plasma and the density of the letters in the plasma that are there. This is related to the sun, how much the sun means the radiation to this layer. And there is also the 11-year sunspot cycle. The sunspots increase and decreases with the 11-year cycle. And right now we are in the bottom valley of the 25th sunspot cycle since the human started measuring it. And then the propagation is the worst possible now, which is good because anything we develop now will work in the next years much better. But by current ITU definition, HF goes from 3 to 30 MHz. So this is more easy to understand drawing. If you want to reach a much faraway place, you can use a sky wave instead of ground wave, line of sight. That's usually what we use in higher frequencies. In very low frequencies, you use ground wave. Then the Earth works as a wave guide to the signal. But in HF, no. In HF, we use this type of transmission propagation that's called sky wave. Brief history. Sky wave transmissions in HF were the most important means of communication for places where there are no cables to connect. For example, the imperialist powers to talk to Africa. They use HF and there are a lot of telephony services running on HF in the 30s, 40s, and so on. Till today, HF is widely used in Amazon forests in Brazil and Congo, South Sudan, Central Africa, Republic. It's the only, in the end, communication means. Satellite is expensive for most of the people. Well, the advantage of using satellites is that in HF you can put how much power you want and people will be able to receive a small, shi-t-hf receiver with a small telescopic antenna and satellite. Usually, you have power constraints in the transponder to send the signal back and also the latency of satellite is much, much greater than HF. You're bouncing 100 kilometers only and the satellites are thousands of kilometers high. Well, the clear disadvantage of it is that you lower frequency. So you have less bandwidth, you have bigger antennas. So, yeah. Well, this is a place I worked with some HF installation in Acre. That's this state of Brazil in Amazon. Well, the characteristics of this Amazon is all here, no, the top of this map that shows part of South America, the more light color is Brazil. And, well, there is no roads, there are no trains, nothing. The transportation is by rivers, no electricity also in the rural areas. And, yeah, it's kind of tough place and they really need communication for emergency, for any kind of communication. They miss it because if they get sick and the river is low, they just die. They cannot even call someone. And, yeah. So the other project, Hisomatica now is planning to work, is in Pará. That's the top state of Brazil, yeah, close to Shingu River, thanks, Kit. Yeah, more or less, sorry. More or less in this region, that's called Teja do Meio, that's like the size of Holland and Belgium together, a bit bigger. And the distance from the municipality where you have internet and every other possible connectivity to the communities we work is 500 kilometers. It used to be the biggest municipality in the world and they split it a bit, so now it's in China, but it's the second. And so the links usually is from the forest to the near cities, that's 500 kilometers far. Yeah, this is a typical house. There is the oven, people hunt, so everybody shoots very well. Like this is a typical installation that we did, this was in 2016. It's a common ICON IC78 radio that's used for hams. This is a commercial version that's unlocked. You can transmit in any frequency you want in the HF band. One simple battery connected to a solar panel. Yeah, no big deal here. Well, then in Hisomatica, in the past year we came up with this idea of applying to this challenge that Mozilla launched in the beginning of the year, that's called WINS, Wireless Innovation of a Network Society. There are two challenges. One was for post-disaster situations, and then we proposed this project for this situation, like a disaster or a catastrophe, something like this, how to recover and have a communication system when the communication infrastructure is destroyed. So the idea is to have, this is a remote location where, for example, an isolated community or let's say a place where the disaster happened. So the idea is to put here the HF radio, BTS, Wi-Fi access points, stuff like that, and then you connect to the nearest hop where you have connectivity that you call here Hermes Central, that you connect the HF radio to the network and so on. If anyone wants to stop to ask something, feel free. Okay, the equipment we used in this trial in Mexico, in Oaxaca, was this icon IC7100. The radio indeed is this bottom part, together with the part on top. In the middle is just a wattmeter that also measures the reflected standing wave. And it's also an impedance matches, so if the antenna is not good, you just match it there. Power supply, a small wireless access point, a PC and a BTS that you all know very well. Yeah, this is in one community when we were installing the HF system. Of course we don't take masks in the middle of the forest, so people chop a tree and make the mask with wood. That's the antenna, one simple, that specifically is not exactly a standard dipole. It's a Carolina Widow antenna that has one side, one half of the antenna bigger than the other, so it gives you a little wider bandwidth and there is a balloon in the middle to match the impedance. And this antenna we were using for 80 meter band, like around 3.5 MHz. Well, in the Hermes project we decided, in the first projects I participated, we were using DRM, Digital Radio Mondial, that's a standard for broadcasting, it's not really very good for it, so we decided to change to AirDop, that's a modem developed by Herm Radio people. Can give you up to 2 Kbps, can use with any HF radio. In this trial in Mexico we use 6 MHz, 6.030, something like this. The type of transmission in the HF is NVIS, Near Vertical Incident SkyWave. This means the antenna is close to the ground, the signal goes all upwards, so you have no black zone, you have a continuous coverage of 600 km, depends of how high you put the antenna. In this trial in Mexico we used a standard quarter wavelength dipole antenna, just cut the wires and connect to the radio, no big deal. I wrote this HISOHF connector that I talked to the TNC, these are DROP TNC, and expose an interface to the RCCN, that's a key to talk about, but it's just an I-notified basis, it's put directors input and output directors that you drop a file there, it transmits the file, when a file arrives it writes there. The service is a key to talk more about, but we enabled SMS and voice messages, the voice messages we encoded with Kodak 2 in 1.3 kbps, that was the lowest bit rate that Kodak 2 provides a understandable audio, less than this is kind of very bad, it's shit, this is what works more or less well, and some characteristics of the modding, like you can choose different bandwidths, force, different bandwidths, otherwise it chooses adaptively depending on the propagation characteristics, you have two mode, the one ARC that's automatic request, it's a connected mode, so if the other end doesn't receive a package or a frame it asks the other end send me again, and there is a broadcast mode that's called fact, that you just set the how robust you want the signal, the fact level, the decoding rate, and also how many retransmit times you want, just like any data carousel protocol in any broadcasting system, so we just want to retransmit 5, 10 times and broadcast it, the modulations are this one, 4ps key, 8ps key, 128 carriers, 25 to 167 baud, that gives us 28 up to 2.296 kbps, the modding chooses what works, if the frame starts to be received in the other end with error, it goes increasing the robustness and decreasing the bit rate, and this works pretty well, just the same time you started doing trials, this AIRDOP developers released an experimental FDM mode, of course we didn't use it because it was just released in this previous mode, like it was working fine, but with FDM you can get more or less twice the bit rate, so you can get more or less a 2 bit per second per hertz of bit rate density per hertz, which is good, things are getting better, people are really evolving, well the AIRDOP communicates with the radio using standard AUSA to send the signal in and out, and to key and unkey the radio, serial port, TTIPSLONG USB 0, this radio we bought has a USB cable that exposes two serial ports and the AUSA in out device, one serial port is basically for key and unkey the radio, basic stuff, and the other serial port you can control all the parameters of the radio power, frequency and everything, that we didn't use the second serial or just tuning the radio and the frequency we were using, that's it, this is an example of the command line, like you put a TCP port, that TCP and TCP plus one ports that used to connect to the TNC and the other port for the payload, and you put the commands to send to the serial to key the radio, to unkey the radio, and these are the AUSA device names that I set it up, AUSA RC here, because it uses 12,000 hertz sample rate, the modding, so some sound cards, and specifically the radio sound card was 48, and AUSA was doing some shit, and then I created this to do conversion in software, and it worked pretty well. Just for the sake of completeness, some comments of the TNC, like there is the arc call that's to a connected connection, the buffer that the modding says to you how much data is inside the TNC, still to be transmitted, initialize, listen, you said listen on, listen off to say if you want to receive connections or not, like protocol mode, that's the arc or the fact, and this is what you send to the TNC, this is what the TNC sends to us, like buffer to say how much is still in internal memory of the TNC not sent, connected, disconnected when you manage to connect, PTT when the TNC keys the radio and keys the radio, not much complex, I just wrote software to use this, and well, what really the user in this, that's using the service sees, no, so keep, do you want to talk a bit about... Yeah, yeah, quickly, so this is our seriously, you know, complex protocol that we invented to pass messages from one system, Hermes central to Hermes remote, which was done as Rafael just explained by dropping a file into a folder on one system, which the HF connector just picks up and says, okay, I need to transmit this, and so it's like some kind of R-sync or something that just takes a long time. Whatever file you drop into a folder on one system will eventually appear on the other system. So we decided to encode SMS by calling the file name, message with the SMPP sequence number, and the file contents of the SMS was simply the first line, the source number, the second line, the destination number, and then the message text after that, and our call is a call with the UUID generated by the PBX, and again, just the source and destination phone numbers in the file name, and the payload is the codec 2 payload, so it's no big deal, just we just keep things really, really simple, so just don't need to put any other information in a non-text file, we just encoded the metadata in the file name. But how does the actual transfer of the file work? I mean, as you said, you put together some files of the name or whatever, but how do you transfer the actual file over there? I notify interface when the connector sees the file is there, read the file, write it in the ring buffer, the other thread reads it, and write to the TNC slowly because the TNC just overflows if you write too much data, and the TNC does the sender station, they receive station, it can do, it does in this case automatically, so if both stations are trying to send, it first ends, let's say, one station finished to send the data, then the holes of the station change for transmitting station to receive station automatically, and the other station, if there is some data to send, for example, the file sends, the modem is like a dump pipe, the TNC, everything you write. What I'm saying, everyone, I'm interested in this exactly, the protocol that you're using to convert files into that dump type. Yeah, it's like what I do indeed, I write in the beginning of the file, the size of the file name, the file name, and then the size of the payload, no, no, no, I just write the size of the payload, and then I have the file name and the payload just after and then pump this, no, I have two sizes, the file name and the payload, and then just write this, the modem takes care that everything arrives perfect or doesn't, it's like a connected mode, so it's a dump pipe, you write whatever you want, and I did in a way that you just write to the ring buffer whatever you want, and the modem will transmit, it doesn't matter if it's a file can be a stream or anything, like, yeah. So I said, kind of quick rundown in case it wasn't clear what happens on the SMS's are rather more simple, but in just in terms of this asynchronous phone call idea, was that you have a user in the remote community connected to the GSM network and they want to call somebody, so they call the normal PSTN number and the PBX in the local community picks up and plays back a message to them which says, please record your message for the destination, they record the message and then they hang up. We have, meanwhile we've recorded this message to disk, we encode it with the codec 2 and we drop it into the RISO HF connector spool, which then transfers to the other side over the HF link, eventually it gets there, the other side through the iNotify mechanism sees the file, looks at it, figures out, okay, this is from this number to this other number, it calls that number on the PSTN and plays back to the callee when they pick up and things saying, good afternoon or whatever, you have a new message from a user on the Hermes emergency network, please stand by and listen to the message and then we play back the message to them and then we drop them into an IVR that says, if you're happy, you receive the message, press one, if you would like to hear it again, press two, if you would like to reply, press three, if they press three, we record their message, we drop it into the spool, we encode it with codec 2 and we send it back to the other person and basically rinse and repeat until the communication is complete. So do you have a max time of recording, let's say like 30 seconds or something like that? I mean that would be really easy to implement, I didn't implement it in the test case, I don't know, I mean we were sort of looking at that with codec 2, at the bit weight we were using it should have been more or less real time to transfer the files but it didn't really end up like that, it was about 2 by or something, so I guess the idea is that people should be informed about using the network that they would say try to keep your messages short because the longer you make your message the more difficult it is to get it through. I mean it's actually a good idea because every time more I have friends just actually instead of just messaging me they actually use like messaging applications to actually record voice over two minutes and then I have to listen for it so actually it looks like quite real case scenario. I think we limited, we limited the sizing 15 seconds. Sure. I mean the idea of recording and then sending and whatever, I mean I see that actually people use it so yeah it makes sense. Yeah I know what you mean. So yeah in the same token with SMS when somebody sends an SMS it just drops it into the spool file, gets transferred over and gets transferred to the other system and vice versa. What is the usual delay for say 10 second message? I mean how long does it take for it to be transmitted in the real world and arrive at the receiver? I think it was about 2 by in our real world test, no? Was it like for a 10 second file of 20 seconds on the HF? Maybe 2.5 something like that. Less than three times no? In the real world testing in one station in the mountains one in Oaxaca was like two times the size of the message but yeah we were using an old version of AirDop that the betrays much lower than the current version so I think the same size of the messaging time would be the time to transmit it more or less. Yeah I wanted to demo it but I don't actually have a radios with USB or serial PTT and so I was thinking which is using the sound card which is where you might have heard AirDop screaming in the back there a little while ago but I decided to abandon the idea of doing a demo in the time span we had. But yeah it's easy to play with it yourself. So how does the system understand that this file was already transmitted and this one is new and this one is old? It's a notify, like I just read the file and write in the ring buffer if there is another one I get and write in the ring buffer and in the other end the connector just keeps reading from the air, writes the ring buffer then there is an adapter that keeps reading for the ring buffer and in case a file adapter that there is the syntax of file name and payload then I get the file name create the file and the payload I write to the file. So how does it survive the restart or do you remove the file after it's transmitted or? After it's transmitted I remove the file exactly just this. Well it's not complete yet there are some corner cases certainly that it's a prototype like if we put in production some things need to be more care, give more care to. So I'm not sure if you're using UECP for that kind of transfer because that's exactly what it's designed for. The problem with existing UECP protocols is that they do basically segmentation and retransmission and all that stuff which the T and C is doing but I think what should be rather easy is to, so UECP is specified both over analog links where you do X modem or Z modem whatever with all the retransmissions but it's also a transparent mode which is used if you use UECP over TCP normally. The advantage of using UECP would be that you have all this infrastructure around so you basically can execute remote commands so you don't have to say oh this is whatever a file for voice or something like that but basically the sender basically says well execute this shell command on the remote node after the file has been received and it gets executed and you have also the notion of batches and compression and so on and so on so might be an idea if you haven't looked into it to look into some synergies there because UECP has been out since I don't know 30, 40 years. Another option was to use sound modem, the good outside sound modem from Tonsoyer and UECP but in the end I realized the modem itself not UECP. This was kind of buggy and I think it got bit rough since 94 so we went to something new that doesn't yet talk to UECP but I think if you go to production with this that's the idea. I think certainly we need to evolve it a bit. I mean I need to see what would be the advantage of putting UECP on top. I think for sure that we should look at what 40 or more years of protocol development has got for us. Some of the things that this is sort of a precursor to ideas of how to do some kind of interaction of communication over HF link where you obviously can't run TCP or anything like that so I mean my idea with a little bit of this is to try to figure out some sort of IRC gateway or something like that where you might actually maybe you don't want to be sending all messages from a busy channel but you could have some kind of you could send mentions or you could just have a connection to an IRC channel with a rate limit on it so you would stop forwarding in the case you got too much and that would allow people in an emergency situation to be connected to a lot of other people without having to implement that multi-person infrastructure yourself. You could be remotely connected to an IRC channel that's coordinating from far away just going to have quite a lot of latency on your conversations. Yeah but this is very important we are developing the service to you because you just cannot run normal internet on this kind of technology so we need also to provide the services or at least ask them what they want and this idea of IRC is good. More slides. So some links and we expired the time already a bit. I have some slides that I want to show if you think it's okay like some HF broadcasting facilities in Brazil that the government owns and well HF broadcasts are being shut down all around the world and what they discovered just testing is that this incredible huge HF antennas can be used for two-way communication and so I'm trying to pass more or less fast. We did a test using a HRS 440 that's a curtain array antenna for dipoles with reflector of 16 dB of gain running DRM that's what we wanted to run to have a bit higher bit rate. You can see the signal here, the preamplifier, the wattmeter and yeah in that frequency. So our simple setup like a lot of things, the fun not to burn everything, power supply, DRM radio. This is the building. In the back you can see one antenna there are 18. This is the transmission line and balance open transmission lines. I think there are 300 ohms. Transmission line in the back one antenna. This is looking to the below the antenna. You can see the dipoles, the reflectors. It's me in front of the antenna. It's not a small antenna. This is one of the antennas. Where we want to go with this in the end like we would like us to have our dog doesn't have it but we have a routing system that if third world war comes and if North Korea should so the satellites would like to have a proper network that we can connect anyone in the world using HF. Right now the radios only can transmit or receive so full duplex would be nice. Like wider bandwidth like the radios are limited because they are made for voice so you have more or less a low pass filter around 3 kHz so this is what limits our bandwidth. Military radio can do a lot more in HF. Like I saw the demonstration of military radio from Brasilia to Manaus like 2.5 thousand kilometers they managed a stable connection of 96 kilobit per second which is pretty fair. These are all open standards by NATO. This standard is the 2GHF. Though I could not find much free software that implements these NATO standards but I don't know if it's really worth because they're fairly complex. I don't think we need all that complexity. It's just like you do in CISMACON to implement all the 3GPPs tech. I think it's too much to implement the NATO standards which is something simpler but still with the functionality that can be useful. Yeah and the thing we want to do sooner than later is just have a phone call, GSM phone call over HF but as you cannot have bi-direction communication like some use a VOX activity detection and this continuous transmission from GSM as a hint to have a person in the community to have a live phone call and maybe implement some way the PTT or just use VOX activity for a person to really dial to someone and complete the call using HF as back call. Using these radios you could see. But now you're talking about the digitally encoded voice or you're talking about using the analog HF channel to backroll the voice? It could be anyone. It could be analog or digital. In the end I'm thinking digital because the quality will be better. Sometimes not but yeah usually digital but it could be analog like it doesn't matter. It could be SSB. Just the signaling needs to be digital. But this is what I think next step it would be fun to have. So do you have an idea of which kind of latency do you get in the setup? Do you have one of your towns let's say or like from one side to the other? Approximately. It's very fast in the end. Almost line of sight just the signal goes up a bit. Yeah but that's why so I mean it's going up and down. The problem is the Doppler because sometimes the F1 layer reflects a bit of your signal. The F2 layer reflects a bit. So you have multi-path. You have a lot of problems like this but with digital modulation in these days you can cope with OFDM systems. You can easily take these problems out. The latency I cannot tell you exactly. We didn't measure and didn't care in the end. Yeah I mean I was just wondering if it's like similar to satellite. No no. It's more like a landline. I mean it's slightly long. So if you think you're talking about what it's like 500 km distance but you need to go in addition you can go up to 100 km down. I mean it's still very little. Yeah you can have Doppler of till 5 milliseconds because of the signal boss in different heights. The latency is lower. It's much lower than the satellite. Do you want to say something about the problems of the full duplex radios and the duty cycle of... Yeah like the problems ahead is like the radios got extremely hot because they are not really made for full duty cycle transmissions. So we had to decrease a bit the power. They are made for voice and voice like SSB is carrier suppressed so between each full name there is zero power being transmitted and in any digital system, let's say, but this OFDM or single carrier digital system you always have like full power and these transceivers are not really made for it. Also the battery system that is designed for voice cannot do digital because they run out very fast. The radio sucks much more electricity. Like small problems that you only know when you do it and stuff like that and the problem of doing full duplex I heard people like asking people oh what if I make a radio transmitting a frequency shift up and down just like GSL. You cannot have a practical cavity filter on this size so you need to have other ways of separation and people say oh just put the antenna 100 meters far from the other and then you have your separation. Even these military radios, they are not full duplex. They can get very high throughput because they will spread spectra in the HF and so on but still they are not full duplex. They are just one way and it would be good to have two way but there are some problems that doesn't seem that easy but I'm not an expert so yeah we are thinking about how to do it. I'm just wondering maybe it's a bit of a dumb question. Saw you signing in front of a not small antenna Do you require that or do both sides have to be this big or what is the relation? No, the point is that antenna is first that they are there and they want to scrap that thing. I showed previously the same antenna zero. The point is that with that antenna, with that gain we can make a wireless router that can cover the whole Brazil. You have antennas pointed to all directions. The idea in the past was to do jamming in Voice of Russia Radio-Humania International from the communists. So you had 600 kilowatt transmitters with antennas pointed to all directions to be able to jam radios from the communist side but today the context is not this anymore. So the antenna in the received side is much more sensible so one transmitting with a simple antenna to receive much better with this huge antenna and to transmit like you have huge gain so you can go much far away. So this is why you would like to use this antenna. So just to recap, like the setup with the solar panel and the small radio on the desk and the small gray capsule-like antenna that's enough to transmit back over 500 kilometers? Easily. Like this antenna was meant to transmit 10,000 kilometers like to transmit to Africa, to Asia, the big one. No, no, no, this one, no, no, this one, no. Let me just throw it back. This one in the top. That's a simple antenna, like can be a dipole. With this one, like you'll be able to do long paths but you'll be received a very, very attenuated signal in the end. So the idea to use these big antennas is to really... The problem with the big antennas is that they have a very low angle of attack so they are really made to very long paths. So one thing we are considering is changing a bit the radiation pattern instead of sending close to the horizontal send a bit higher in order to cover the country and not to be received in the U.S. better than 100 kilometers. You have really a black zone. Like if you go through 100 kilometers far you cannot receive hydrogenation of the Amazon and then after 800, 700 kilometers you start to receive the radiation because you are in the hop of the signal. You are in the jump, so... Yeah. So you mentioned this particular antenna that we're seeing now. It was made for quite long distance but it has high attenuation. Which antenna? No, this is a single unitary gain, one-quarter wave. This is exactly not one-quarter but imagine it's a single dipole. It's a single one-quarter wavelength dipole like the theory is easy to understand. People usually install it as inverted V because if you install a strictly horizontal dipole you have nulls in transmission in the antenna so people usually want to talk with everybody so they install instead of like these things they install inverted V configuration and then you can have a more or less omnidirectional pattern so you have a contiguous coverage adjacent to the antenna. This is like... And it's simple to do, it's just a wire. You mentioned the received signal has quite a bit of attenuation as the distance grows, right? And what kind of dB values would that be, do you know? I don't know, to be honest. Okay, putting numbers, if you have less than 40, 30 watts for voice that you need 6 dB at least of SNR you'll not be received. Using a quarter wave antenna in an horizontal configuration typically all the radios, all the HUM radio have 100 watts more or less but this is for SSP, if you put our FDM of course you'll not get out of it 100 watts but I don't know, but this is easy to find because it was being modelled since the 30s more or less and the point is that you don't really need to model because there are ionosomes in many parts of the world that keeps sending vertical signals in one frequency up and down in different frequencies in the end. So the most important parameter you need to know from the HF is they move the maximum usable frequency that's the maximum usable frequency that the ionosphere is bouncing and the attenuation you can get experimental live data from these ionosomes so you know the attenuation from the live data you get from these beacons that are being thrown to the ionosphere I don't know the numbers but you can get live these and they are prediction software also but the live reports are better, of course.